TM1 Forum

I have been working as EP 8.3 developer for last few years and just recently started learning TM1 9.5

Can someone tell me how do we manage the below situations in TM1:-

1. In EP we have the option of migrating the model from QA to Prod without data so that the data present in Prod. model remains intact unless it is a destructive sync. How can we do this in TM1? Also in EP we can either copy the database or use the export wizard. What are the different available options in TM1?


2. How do we create the break-back functionality in TM1. Is data spreading the only way to do this?

Thanks

Much as I love working with TM1 there are definitely some areas where EP beats it hands down. One very weak area is deployment. No tools are provided (there are some 3rd party products) and you need to have a very good understanding of the contents of your server to do it safely. The key is the object datestamp - TM1 does update this properly. However there are a number of objects which are updated all the time eg performance monitor cubes. Take lots of snapshots and be prepared to do it more than once.
On breakback, well spreading was written by Applix as a counter to breakback. As you have spotted, it's not really the same. If you go to 9.5+ Contributor, you'll see something more like EP breakback, but it uses the spreading code plus sandboxes behind the scenes.
Over time, more of the EP features are supposed to move into TM1.

David,
Thanks for the response.

I don't fully understand the first part about the change migration or deployment from QA to Prod.

Are you saying that unlike Enterprise Planning, where we can just create a deployment in QA and import it in Prod., in TM1 we will need to make all the changes manually in Prod. if we don't want to loose the data already present in there?


I found another extremely tedious method of doing this by exporting the cube data to cma files and once the server folder has been copied from QA to Prod, use TI process to import the data into all cubes from the cma file that were previously created. This is no way an elegant method either.

I want to know which of these two methods is the commonly used by TM1 administrators/developers?

Thanks.

jbcraigs wrote:I found another extremely tedious method of doing this by exporting the cube data to cma files and once the server folder has been copied from QA to Prod, use TI process to import the data into all cubes from the cma file that were previously created. This is no way an elegant method either.

Like David told you, there are no built-in tools in TM1 to help you with migrations. You really have two options:

1) Go to the data folder of the source folder, identify which objects need to be moved, move them to the destination server, and recycle the TM1 service, or
2) Mirror all the changes you've done in the source instance to the destination instance, or
3) Some combination of 1 and 2.

In real life, the more you do of 1) and the less you do of 2), the quicker the migration goes. However, it is more dangerous because of the potential for lost data. I always do as much of 1) that I can because I have a good understanding of the ramifications and know what to do to not lose data. If you are inexperienced I would lean towards doing more of 2). In either case it always is a good idea to make a backup of your destination servers TM1 data directory in case you hose something up.

If you are really worried about losing data during migration I would dump every cube in the destination server to a flat file and then create TI processes that reload the data back in the cubes (make sure to have a step that zeroes out the cube before loading). Then before every migration, run these processes so you have the raw data. After migrating the objects you can run these processes which will empty and reload the cubes. As long as you use the same file names each time these processes can be re-used at will. Keep in mind that if your dimesions have changed as part of your migration then some of your data may not load. What I mean by that is if you have amounts in account "10000" in production and part of your migration is to close that account and remove it from TM1, when you go back to load the data from your flat files that record will reject when trying to reload. You will need to do extra steps before extracting data if your intent is to merge history as part of your migration.

Thanks a lot Tomok. This is the information I was looking for.